基于自适应投影增强的压缩感知CT图像重建

Compressed Sensing CT Image Reconstruction Based on Adaptive Enhanced Projection

针对投影数据受噪声影响严重、被检测工件体积大获取投影数据不完全等导致重建质量差的问题,提出一种自适应反锐化掩模算法对某发动机模拟件投影数据均匀抽取60个角度投影进行预处理,根据其边缘信息设定阈值进行自适应去噪并对高频信息乘以自适应增益增强,采用TV最小化约束的代数重建(ART)算法重建.将改进算法与经典ART、经典反锐化掩模技术处理投影以及投影未处理的POCS-TVM算法进行对比分析,实验结果表明,该方法能够平滑噪声,提高对比度,进而提升不完全投影重建图像质量.重建同样大小的图像,相对于全角度投影迭代一次时间100 s,60个角度投影迭代一次18.7 s,时间缩短近6倍,大量节省重建时间.

In order to improve the reconstruction quality caused by noise seriously and incomplete projection data obtained from the large volume of the detected workpiece,an adaptive unsharpening mask algorithm is proposed in this study to pre-process the projection data of an engine simulator extracted 60 angle projections uniformly from.According to the projection edge information,the threshold is set for adaptive denoising,and the high frequency information is multiplied by adaptive gain enhancement.The ART algorithm with TV minimization constraint is used for reconstruction.The improved algorithm is compared with classical ART,classical unsharpening mask technology for projection processing and POCS-TVM algorithm for unprocessed projection.The experimental results show that this method can smooth noise,improve contrast,and then improve the quality of incomplete projection reconstruction image.Reconstruct the same size image,the full-angle projection is iterated once for 100 s,and the 60 angle projections are iterated once for 18.7 s,the time is shortened by nearly 6 times,and the reconstruction time is saved a lot.